In Greek mythology, Ariadne, the daughter of King Minos of Crete, gives Theseus the thread to find his way out of the Minotaur’s labyrinth. Likewise, the founders of Ariadne Genomics are hoping their ontology-based natural language processing technology will provide a means for researchers to navigate the maze-like landscape of biological data and literature.
But the year-old company is undertaking its own journey into the perilous commercial bioinformatics market with relatively little guidance. The company’s four founders — all former InforMax employees — are funding the start-up entirely on their own and are planning on staying small and focused in order to stay successful. A total of six employees work at the company’s Rockville, Md., headquarters, with some development work outsourced to a team of programmers in Moscow, said co-founder and acting president Ilya Mazo.
“You can do pretty well in bioinformatics if you stay small,” said Mazo, citing companies like DNA Star, Silicon Genetics, and GeneCodes as examples. “They have been around for a long time and have done pretty well. You don’t have to grow to more than $10 million per year.”
At InforMax, Ariadne’s founders witnessed the downside of rapid growth first-hand. Mazo left his position as scientific project manager at InforMax in December 2001 and founded Ariadne shortly afterwards. Former colleagues Nikolai Daraselia, Alexander Nikitin, and Sergei Egorov soon followed. So far, Ariadne has signed five customers for its software and also plans to apply for SBIR grants for additional revenue. A “small contract” for genome annotation work for NCBI also helps pay the bills.
Ariadne’s flagship product, Pathway Studio, sells for only “a few thousand dollars,” according to Mazo, and can be downloaded from the company’s website (www.ariadnegenomics.com). The application is based on the company’s MedScan natural language processing algorithm for extracting information about protein pathways from Medline abstracts. The package also offers a drawing tool that allows users to graphically represent pathways with information from the literature, as well as a database of over 100,000 interactions between 15,000 proteins, cell processes, and small molecules.
Mazo acknowledged the handful of other companies are developing similar technologies. PubGene, for example, also sells text-mining technology with pathway visualizations, but Mazo pointed out that that while PubGene only “looks for word co-occurrences, our technology…understands the role of each word in a sentence and builds the grammatical structure of each sentence.”
One benefit of this approach, according to Mazo, is that while the current offering is focused on pathway analysis, the technology uses a context-free grammar so “it can easily extend to any other area of interest” to extract information about links between genes and diseases, for example. Mazo said that the company has designed a specialized ontology for pathways, but other ontologies could be created to tailor the technology to any biological relationships of interest.
Mazo said he’s pleased with the company’s progress, but the company’s first year wasn’t all smooth sailing: “Of course, we’ve done everything wrong,” he laughed, “but we survived the first year, so I’m enthusiastic.”
If Mazo and his team stay on course, perhaps Ariadne Genomics will experience a better fate than some of its bioinformatics predecessors (as well as its mythological namesake, who was abandoned by her lover Theseus on the island of Naxos).